Hmm! it appears we have A4 (316) SS failing on a sulphate crag? It’s time to measure stuff.
Many thanks to Simon Alden for the sample and photographs.
This story comes from the sea cliffs of Malta. Whilst I don’t have significant geochemical sampling for Malta, I have every reason to believe the Mediterranean location will encourage the presence of sulphate. See this earlier post for a detailed explanation. To date, I have one positive identification of a 304 bolt from Malta that has failed under the attack of sulphate reducing bacteria (SRB), so I guess another is possible…. but 316?
Some ten years ago Simon Alden installed a couple of 316 SS twist bolts from Bolt Products using Hilti RE500. This route on Gozo is rarely climbed, and it was only during a recent visit that Simon was surprised to see both anchors were clearly cracked.
Simon cut the bolts off and sent me the left hand one. My first action on receiving it was to check attraction to my strong, rare-earth magnet. This is an important test, and I did it even before taking the sample out of the bag. It seemed strongly attracted to the magnet.
We can get a lot more exacting than “seems strongly attracted to a magnet”. To this end, I have developed a simple technique where I measure the force required to pull a tiny rare-earth magnet away from a contact point on the surface of the bolt. By various empirical means, all imperfect, I have arrived at the approximate relationship between that force and the percentage strain-induced α-martensite at the contact point.
According to my rough calibration, we are seeing 58% martensite here, and, based on the rate of diffusion of atomic hydrogen through austenitic steel containing different fractions of martensite, I wouldn’t expect that bolt to survive in a sulphate environment. This prediction accords with the level of damage we see.
Having decided that this bolt couldn’t be 316, a bit of analytical chemistry was called for. This is what I found. Methodology is here.
| element | composition (%) | error (%) |
| Nickel | 8.4 | +/- 0.2 |
| Chromium | 18.1 | +/- 0.1 |
| Molybdenum | < 0.1 |
My conclusion is that we are looking at a 304 failure, not a 316 failure.
Evidence for SRB attack on 304 anchors located within the Mediterranean, and along the Portuguese coast, is gradually accumulating. Failure modes that were originally attributed to SCC (stress corrosion cracking) are beginning to be revealed as SSC (sulphide stress cracking).
What evidence can we find for that mechanism?
In cross-section the stress cracking is obvious.
The real test for an SSC mechanism is to demonstrate the presence of sulphide.
This was easily done using the Iodine-Azide spot test.
A further tell-tale sign of SRB releasing sulphide is the formation of the mineral greigite, Fe3S4. The deposition of this particular iron sulphide is likely to be favoured by pH and oxygen level at some point between the inside of the bolt and the free atmosphere. See this post for a discussion of the chemistry. It is quite distinctive with iridescent, octahedral/cubic crystal facets. The picture below shows the greigite deposit I found.
If we take a close look at a polished section of the cracked region, we find the metal to be fragmented and snapped on the microscale. The grain size, not observable in the photos, is of the order of 10um to 20um, whereas we are seeing fractures occurring on a geometric grid with a spacing closer to 1um. This is characteristic of hydrogen embrittlement where, according to the HELP hypothesis, micro-voiding is initiated at the intersection of the slip planes of the lattice. I have a 101 introduction to the subject in this post.
So, we observe a ductile material rendered totally brittle at the microscale. It is hard to imagine that such a tough alloy as 304 would powder to dust at the points of maximum stress, yet that is what the photo below reveals. The powdered metal at the fracture has been caught up in the epoxy encapsulant that I use to embed the specimen under vacuum.
Conclusion:
So, we see once again the attack of SRB on 304 rock anchors. The observed stress fractures are SSC not SCC and beneath the surface the hydrogen embrittlement that enables this flavor of stress cracking is obvious.
The challenge still stands: Bring me a sample of a 316 anchor that has failed through SRB attack.
5 replies on “Is this a 316 failure?”
Hi Dave. Thanks for another interesting post. I cannot avoid but think about the bolts from Laguna. At a layman’s eye they look very similar, except that there they were more yellowish all around instead of shiny. I’m not sure there were visible cracks as well.
Best wishes
I owe you an apology for being so slow with that sample. I’m just in the process of preparing a section for microscopy.
It doesn’t look like SRB attack and doesn’t test positive for sulphide. I was thinking it was more likely a metallurgical problem, but it is not hard for 304, so over-hardening looks unlikely.
Let’s see what I can see in the microstructure. I suspect we have something quite different here.
You do not owe me anything! No worries. Thanks for the hard work!
Hi Dave. I was curious about your magnetic balance technique. I understand that it is an “imperfect” calculation, however I am wondering what formula you use to determine the percentage of the martensite.
BTW I am Climbing Taiwan’s wife…so going down these facinating holes is essential especially since we are having SS products made. Understanding the relationship between cold-working, the nickel content, martinesite and what actually is happening to the atoms is amazing. We’ve been comparing some “name-brand” 316L hangers with some generic 316 ones and been a little surprised that the magnetism of the “name-brand” one is a bit higher. That leads me to conclude that some major manufacturers are not having a full solution or stress relieving annealing after forming, (which can also lead to other stresses as a result?) If I am incorrect in my assumption please let me know.
Sorry for the long comment, but we really do appreciate the time and energy that you have spent in this. Your input is invaluable to our community especially for those of us that live in areas that see SCC and/or SSC.
Hi Andra, please accept my apologies for the slow response. I have been travelling, last month.
Calibration of the magnetic balance technique is very approximate. I work on the principle that a specimen cannot exceed 100%, and by looking a a very large sample of 304/316 components, I assume that the very highest is close to 100% and work backwards from there.
I find the results are very consistent, and comparative measurements are likely to be valid. This works for me in a practical sense since there is not much I can do with the absolute figure anyway except broadly compare it with what other authors have obtained in H-embrittlement studied.
With the exception of XRD I’m sceptical of many of the methods of alpha-martensite estimation. I’ve tried a few and haven’t been impressed by the results. Perhaps I need to try harder. Have a look at this paper for a thorough attempt to get at alpha-martensite calibration.
https://www.researchgate.net/publication/233708990_Comparison_of_Different_Methods_for_Measuring_Strain_Induced_Martensite_Content_in_Austenitic_Steels
I’m very happy to see you guys determinedly breaking stuff. We are just talking if we do not actually make measurements. The trick is to tie the results back to the mechanical model. Evidence that points to the robustness of this model, is an indication that a robust QC system can be established.
In an ideal world, you should see no evidence for alpha-martensite formation in a 316 SS part. To do so would imply the part was cold-worked at less than -20C.
However, all commercial 316 bar is likely to contain around 2% delta-ferrite. This is a deliberate. The annealing of the cast slab at say 1050C is cut short so that some delta-ferrite remains as an insurance against cold-cracking following welding.
My guess is that you may well be seeing difference in residual delta-ferrite. I normally see levels of “apparent” alpha-martensite of 1% to 3% in commercial 316 SS bar.
Is delta-ferrite a potential source of hydrogen attack? I think that has to be so given it has the bcc structure that makes alpha-martensite vulnerable. However, I believe the situation is more complicated than this, and I have a long term project to attempt to measure hydrogen binding energies in say strain-induced martensite versus weld-induced delta ferrite.